This proposal describes a novel and integrative Network Biology approach to identifying specific molecular patterns that can define and distinguish different cancers by their specific "Cue-Signal- Response (CSR)" profile. These cancer-specific molecular patterns can be used to identify new cancer therapeutics or predict responders to existing cancer therapies, both of which are of high commercial value. The focus of this proposal is the ErbB signal transduction network's activation of ERK and AKT, molecules from which physiological responses of clinical relevance (Response) can be inferred. An extensive multiplexed antibody microarray system is used to simultaneously quantify cancer-relevant molecules; EGFR, ErbB2, ErbBS, ErbB4, EGF, heregulin, amphiregulin, betacellulin, and TGF-alpha (termed the "Cue"), in addition to ERK and AKT. Computational Biology is then applied to quantitatively predict the molecular signature response of the cancer cell to each specific cue. The overall result is a highly predictive computational model, which based on the quantitative expression profile of the "Cue", allows for the prediction of ERK and AKT phosphorylation, and hence cellular response. Because of the mechanistic basis of the model, the response to inhibitor(s) of any of the kinases in the signaling network (molecular signature of a cancer) may be predicted. Preliminary data show that such an approach can indeed be predictive across multiple cancer cell lines. And thus, it is anticipated that the proposed approach can become a robust diagnostic tool to predict the effect of a specific cancer treatment, such as EGFR or ErbB2 inhibitors.